18

Iranian Journal of Materials Science & Engineering Vol. 8, Number 2, Spring 2011

1. INTRODUCTION:

In the Pidgeon Process, magnesium metal is

extracted from calcined dolomite under vacuum and

at high temperatures using ferrosilicon as a reducing

agent [1]. In this process, the finely crushed dolomite

is feed into kilns where it is calcined. The calcined

dolomite is then pulverized in a mill prior to mixing

with finely ground ferrosilicon. After weighing and

homogenizing the fine calcined dolomite and

ferrosilicon, the mixture is briquetted. Briquettes are

charged in a retort and put in the reduction furnace.

The reduction operation is a batch process releasing

magnesium in vapor form, which condenses in the

cooled end of the retort outside furnace wall. After

removal from the furnace, the magnesium “crown”

is taken off the sleeves [1].

Approximately, 80% of the world demand for

magnesium is currently supplied by China and

nearly 95% of the primary magnesium output of

China is produced using the Pidgeon process

mainly due to low labor and energy costs and lax

environmental act [2, 3, 4]. The main scope of this

research is to characterize Asian Abe-Garm

dolomite ore and its technical evaluation for

magnesium metal extraction in the Pidgeon-type

reactor.

In spite of enormous dolomite resources, two

ferrosilicon manufacturers and relatively cheap

energy, magnesium is not being produced in Iran.

The Iranian market conditions are mainly in favor of

the Pidgeon process, over the electrolytic [5].

Extensive review, fieldwork, sampling and

mineralogical and chemical analyses of major

dolomite resources in Iran [5] demonstrate that Asian

Abe-Garm dolomite is one of the most suitable

dolomite ore in Iran. It is located in area with the

suitable industrial infrastructures (Qazvin Province)

for future development of an Mg plant [5, 6].

In the present study, Asian Abe-Garm mine

dolomite ore was characterize using optical

mineralogy in Tehran Tarbiat Moallem

University, XRF and XRD analysis in Iranian

mineral processing research center (IMPRC).

The calcining and the thermal reduction

testworks using Semnan ferrosilicon were carried

out in Mintek Laboratories in South Africa.

2. EXPERIMENTAL PROCEDURE

Asian Abe-Garm dolomite mine is located in

85 km southwest of the Qazvin city, Qazvin

MAGNESIUM PRODUCTION FROM ASIAN ABE-GARM DOLOMITEIN PIDGEON-TYPE REACTOR

B. Mehrabi1, M. Abdellatif2 and F. Masoudi3

* f_masoudi@sbu.ac.ir

Received: December 2010 Accepted: April 2011

1 Geology Department, Tehran Tarbiat Moallem University, Tehran, Iran2 Mintek Co., Randburg, 2125, South Africa3 Faculty of Earth Sciences, Shahid Beheshti University, Tehran, Iran

Abstract: Ore mineral characterization and various experimental testwork were carried out on Asian Abe-Garm

dolomite, Qazvin province, Iran. The testwork consisted of calcining, chemical characterization, LOI determination,

and reduction tests on the calcined dolomite (doloma), using Semnan ferrosilicon. Calcining of dolomite sample was

carried out at about 1400 ºC in order to remove the contained CO2, moisture, and other easily volatilised impurities.

The doloma was milled, thoroughly mixed with 21% Semnan ferrosilicon and briquetted in hand press applying 30

MPa pressure. The briquettes were heated at 1125-1150 ºC and 500Pa in a Pidgeon-type tube reactor for 10-12 hours

to extract the magnesium. Ferrosilicon addition, relative to doloma, was determined based on the chemical analyses

of the two reactants using Mintek’s Pyrosim software package. Magnesium extraction calculated as 77.97% and Mg

purity of 96.35%. The level of major impurities in the produced magnesium crown is similar to those in the crude metal

production.

Keywords: Mg metal, Pidgeon process, Asian Abe-Garm dolomite, Semnan ferrosilicon, Calcining, Silicothermic

reduction.

Dow

nloa

ded

from

ijm

se.iu

st.a

c.ir

at 1

1:39

IRS

T o

n W

edne

sday

Feb

ruar

y 19

th 2

020

19

Province (Fig. 1). Its estimated reserve is 1.7 Mt

with annual production of 10 kt/y from the

Jurassic Lar formation dolomite. In current

research, samples were collected from Asian

Abe-Garm mine during two session of fieldwork.

Petrography of 60 samples were carried out using

Zeiss Axioplan2 polarized light microscope after

alizarin red and potassium ferricyanide staining

[7] for recognition of dolomite and calcite (Fig.

2). Selected samples were analyzed by Philips

Magic-Pro XRF (Table 1) and Philips Expert-Pro

XRD (Table 2) in IMPRC. Dolomite ore is thin

bedded, beige colored, and exhibit non-planr,

planar S and planar E textures [8] with crystal

size in range of 0.03 to 0.5 mm. The major

impurities are calcite veinlets and fine-grained

quartz crystals (>0.1 mm).

Lloyd M. Pidgeon in Canada pioneered the

silicothermic reduction of calcined dolomite to

metallic magnesium using the abundantly

available dolomite mineral during World War II

[9]. In the Pidgeon process, magnesium metal is

produced from calcined dolomite under vacuum

and at high temperatures using ferrosilicon as a

reducing agent [10]. The Pidgeon process is

currently the most widely used process for the

production of magnesium. This batch process

involves reduction of doloma by ferrosilicon,

carried out at temperature between 1100-1200 °C

under vacuum in a retort, producing magnesium

vapour which is then cooled and collected as a

B. Mehrabi, M. Abdellatif and F. Masoudi

AA 57 AA 56AA 55 AA 54AA 53 AA 52AA 51 AA50

0.04 0.050.050.050.050.020.05 0.05TiO2

n.dn.dn.dn.dn.dn.dn.dn.dAl2O3

0.12 0.070.050.110.120.010.17 n.dFe2O3

31.6032.2533.3240.6427.7033.4030.6333.37CaO

20.4920.5620.2913.3015.3220.3719.0019.87MgO

n.dn.d0.25n.dn.dn.dn.dn.dNa2O

n.dn.d0.02n.dn.d0.01n.dn.dK2O

0.02 0.020.020.020.020.010.04 0.01MnO

0.01 0.02n.d0.010.010.020.01 0.02P2O5

n.dn.dn.dn.dn.dn.dn.dn.dS

5 3 n.dn.dn.d26 n.dRb

66 789813712588109 89Sr

0.86 0.330.000.0011.890.656.09 0.11SiO2

46.9046.7046.0045.9044.9045.5044.0046.60L.O.I

Table 1. Chemical composition of Asian Abe-Garm dolomite samples (as %, Rb and Sr in ppm).

35˚ 45

49˚

15

IRAN

ARAK

1Km

Abe Garm

Asian Abe Garm

Dolomite Mine

Recent alluvium Road

Oligo Miocen Limestone Oligo Miocen Sandstone &Marl

Cretaceous Conglomerate, Sandstone & Limestone

Jurassic Sandstone & Shale Jurassic Dolomite & Limestone

Fig. 1. Simplified geological map of Asian Abe-Garm

dolomite deposit.

Dow

nloa

ded

from

ijm

se.iu

st.a

c.ir

at 1

1:39

IRS

T o

n W

edne

sday

Feb

ruar

y 19

th 2

020

20

Iranian Journal of Materials Science & Engineering Vol. 8, Number 2, Spring 2011

condensate. The reactions of the Pidgeon process

are:

(CaCO3.MgCO3)(s) + Heat = (CaO.MgO)(s) + 2CO2(g)

2 (CaO.MgO)(s) + Si(Fe) = 2 Mg(g) + Ca2SiO4 + Fe

The calcined dolomite and ferrosilicon are

mixed and briquetted to improve the rates of heat

transfer and the solid-state reaction. The major

attractions of the process are its simplicity and

low capital cost; however, the process is also

labor and energy intensive [10].

The experimental testworks were carried out in

the Mintek Laboratories in South Africa. The

calcining of a representative dolomite sample

from the Asian Abe-Garm mine was carried out

in an induction furnace consisted of steel

housing, alumina insulating bricks, and copper

induction coils for four hours at 1400 °C in order

to drive off almost all of the carbon dioxide and

moisture, as well as the easily volatilized

components. A K-type thermocouple was used to

measure the sample temperature and was

positioned just above the magnesia crucible that

contained the sample. After cooling, the mass

loss was measured and sample was taken and

analyzed (Table 3).

Based on industrial practice [11] Mintek’s

Pyrosym software [12] and experimental data

FeSi addition was set as 21%. The Semnan

ferrosilicon and Asian Abe-Garm doloma were

Sample No.

Mineral

AA10 AA13 AA15 AA17 AA50 AA53 AA54 AA57

Dolomite 97.3 74.7 97.1 96.5 98.2 85.0 83.6 98.7

Calcite 2.5 24.9 2.6 3.5 1.8 5.0 15.4 1.3

Quartz 0.2 0.4 0.3 - - 10.0 - -

Table 2. XRD quantitative analyses data.

Dolomite,

mass %

Doloma

mass%

Ferrosilicon

Component Component mass %

MgO 20.90 38.37 Mg 0.03

CaO 31.60 59.71 Ca (ppm) <20

Al2O3 0.01 0.17 Al 1.26

SiO2 0.44 0.91 Si 72.10

Fe2O3 0.12 0.30 Fe 19.11

MnO 0.04 0.03 Mn 0.11

Ni <0.05 <0.05 Ni (ppm) 65

LOI 47.60 <0.05 C 0.08

Table 3. Chemical analyses of Asian Abe-Garm dolomite,

doloma and Semnan ferrosilicon.

Dol

Cal

0.2 mm

Qz Cal

Dol

0.2 mm

A B

Fig 2. Photomicrograph of Asian Abe-Garm dolomite samples after staining, showing calcite fracture filling texture and

quartz grain. A) Coarse grain dolomite and calcite as fracture filling. B) Medium to coarse grain dolomite, patchy calcite

and quartz grain. (Cal=Calcite, Dol=Dolomite, Qz=Quartz)

A

Dol Dol

QzCalCal

0.2 mm

B

0.2 mm

Dow

nloa

ded

from

ijm

se.iu

st.a

c.ir

at 1

1:39

IRS

T o

n W

edne

sday

Feb

ruar

y 19

th 2

020

21

accurately weighed (about 200 g in total), mixed,

and pulverized down to -150 µm in a ball mill.

The mixture was then pressed to produce

briquettes of 30mm in diameter and 15mm thick

using a hydraulic press applying 30MPa pressure.

The final mass of the briquettes was then

recorded prior to being placed inside the retort.

Reduction tests performed on the briquetted

mixture using the retort set-up shown in Fig. 3.

The set up consisted of a 316 stainless steel tube

(100 mm ID and 480 mm long). The tube (retort)

was housed inside an electrically heated furnace

(silicon carbide elements). The furnace insulated

with alumina bricks, provided with a

programmable temperature controller.

The retort contained a gas outlet, which was

connected to a vacuum system. This system

consisted of a vacuum pump, pressure transducer,

pressure and temperature readouts, a shut-off

valve, and argon purge line. Two thermocouples

were used to measure and control the

temperature. The first was placed just outside the

retort, while the second was located inside it to

measure and record the reaction temperature. The

pressure and temperature readings were recorded

throughout the test using a data logger.

After sealing, the retort was pressure-tested by

applying vacuum of 500-700 Pa, switching off

the vacuum pump, and monitoring the pressure

readings. A leakage rate of 0.5 cm3/min, or less,

was considered to be acceptable to commence

with the reduction test. Once the leak test is

successful, the retort was heated up to 700 °C

over a two-hour period. This was followed by a

degassing period of two hours (at 700 °C) in

order to drive off any residual moisture and/or

carbon dioxide that could have been present in

the reactants. For a short period of time, the retort

pressure would increase slowly by 100-300 Pa

during this period, before dropping down to 500-

600 Pa. The temperature was then increased to

1150 °C over a two-hour period. The reduction

reaction was allowed to continue for about 8

hours. The pressure tended to increase slightly

when the temperature approached 1100-1150 °C.

This pressure change was similar to that observed

during the degassing period and lasted for only a

few minutes. The furnace was switched off (as

well the vacuum pump), and the system pressure

was brought up to just above atmospheric by

flowing argon into the retort. Finally, the facility

was allowed to cool down to near room

temperature before opening the retort and

collecting the products.

Both the slag and magnesium crown masses

were measured and recorded, and representative

samples were taken for chemical analysis. In the

commissioning and actual test, a small amount of

powder formed on the side-walls of the retort

(less than 1 gram), particularly near the flange

area. This material was recovered, weighed,

sampled and analyzed separately. Beside actual

test, commissioning test was carried out.

B. Mehrabi, M. Abdellatif and F. Masoudi

Argon gas

supply

Vacuum pump

Transducer Output

Power supply

Data Acquisition

Unit

Argon gas

To

Furnace

Pressure

Transducer

Valve

Furnace

Heating Elements

Lance Briquettes

Retort

Thermocouple

Rotometer

Fibre frax

insulation

Gas

Outlet

Argon gas

To

Retort

Vacuum Rubber

hose

Stainless steel

pipe

Argon gas

supply

Vacuum pump

Transducer Output

Power supply

Data Acquisition

Unit

Argon gas

To

Furnace

Pressure

Transducer

Valve

Furnace

Heating Elements

Lance Briquettes

Retort

Thermocouple

Rotometer

Fibre frax

insulation

Gas

Outlet

Argon gas

To

Retort

Vacuum Rubber

hose

Stainless steel

pipe

Vacuum pump

Transducer Output

Power supply

Data Acquisition

Unit

Argon gas

To

Furnace

Pressure

Transducer

Valve

Furnace

Heating Elements

Lance Briquettes

Retort

Thermocouple

Rotometer

Fibre frax

insulation

Gas

Outlet

Argon gas

To

Retort

Vacuum Rubber

hose

Stainless steel

pipe

Fig. 3. The set up used for experimental work in Mintek Lab. South Africa.

Dow

nloa

ded

from

ijm

se.iu

st.a

c.ir

at 1

1:39

IRS

T o

n W

edne

sday

Feb

ruar

y 19

th 2

020

22

Iranian Journal of Materials Science & Engineering Vol. 8, Number 2, Spring 2011

3. RESULT AND DISCUSSIONS

The chemical composition of Asian Abe-Garm

dolomite, doloma and Semnan ferrosilicon is

presented in Table 3. The Semnan plant

ferrosilicon ranked as Fe-75 wt% Si while it

reanalysed in Mintek showed about 72% silicon

and 19 % iron. The products of silicothermic

reduction test consisted of three distinct phases;

crown magnesium, a white deposit that tended to

stick to the side-walls of the retort (less than a

gram), and the reacted briquettes (slag). Each

product was carefully removed from the retort,

weighed, sampled, and sent to Lab. for analysis.

In Table 4 both iron and silicon are expressed

as oxides. This is not entirely true, as most of the

iron and un-reacted silicon tend to form a FeSi

alloy (residual ferrosilicon). Other metals (Mn,

Ni, etc) tend to concentrate in the FeSi product.

Chemical analysis of crown magnesium

suggests that impurity levels are not significantly

different from those contained in crude

magnesium produced in the Magnetherm or

Mintek Thermal magnesium processes [13, 14].

The magnesium extraction calculated based on

two methods (Table 5). The first is based on the

mass difference between the feed, the slag and

the magnesium content of the crown. The second

approach takes into account the slag mass and its

magnesium content. As such, the later results in

lower magnesium extraction, compared to the

first method. Magnesium extraction calculated

based on the mass of briquette residue (slag) and

its magnesium content is more reliable.

Magnesium extraction defined as:

(Magnesium in feed – Magnesium in briquette

residue)/Magnesium in feed(100%),

where:

Magnesium in feed = Doloma mass*Mg

analysis in doloma

Magnesium in briquette residue = Briquette

residue mass * Mg analysis in Briquette residue

mass

The last column in Table 5 is a preliminary

estimate of the magnesium condensation

efficiency indicates that the calculated value is

very high for the test. Magnesium condensation

defined as:

Slag Mass% Mg Crown Mass%

MgO 8.46 Mg 96.35

CaO 59.36 Al <0.05

Al2O3 0.96 Si 0.05

SiO2 35.14 Ca 2.61

FeO 5.81 Mn 0.06

MnO <0.06 Ni <0.05

Ni <0.05 Fe 0.06

Table 4. Chemical analyses of the slag and Mg crown

(mass %).

Extraction, % Condensation

Efficiency, %

Mass Loss Slag Analysis

86.34 77.94 100

Table 5. Magnesium extraction from Asian Abe-Garm

dolomite.

Fig. 4. XRD pattern of Asian Abe-Garm dolomite, Mg

crown, white precipitated inside flange and slag.

Dow

nloa

ded

from

ijm

se.iu

st.a

c.ir

at 1

1:39

IRS

T o

n W

edne

sday

Feb

ruar

y 19

th 2

020

23

Mg condensation = Magnesium in the

crown/Magnesium extracted *100%

The crown magnesium, slag and white deposit

in side of the retort subjected to XRD (Figure 4).

The crown magnesium was also examined using

Cameca SX-100 EPMA (Figure 5) and TESCAN

SEM (Figure 6). The area covered by crystalline

magnesium is pure while in low crystalline area

which seems formed in the late stage (Figure 6-

C) of condensation there are traces of calcium,

silica and iron (Figure 5). The main impurity in

the crown magnesium is ferrosilicon (Figure 5).

4. CONCLUSION

Chemical and mineralogical analysis of the

Asian Abe-Garm dolomite sample indicates that

it is suitable for magnesium production in a

Pidgeon process. Testworks on Asian Abe-Garm

calcined dolomite and its reduction by Semnan

ferrosilicon in classic Pidgeon-type retort yield a

suitable magnesium metal in terms of magnesium

extraction and crude magnesium metal quality.

Magnesium extraction from Asian Abe-Garm

calcined dolomite is within the expected range.

Magnesium extraction is 77.97% and magnesium

grade is 96.35% in the experiment. In spite of low

Si content of Semnan ferrosilicon (72%)

compared with the industry standard (75%) and

no addition of fluorite as a catalyst to the mixture,

magnesium extraction was acceptable, indicating

the suitability of Asian Abe-Garm dolomite for

Mg metal production. The level of major

impurities in the magnesium crown is similar to

those in the crude metal production.

B. Mehrabi, M. Abdellatif and F. Masoudi

Fig. 5. BSE/EPMA image of crown Mg and chemical composition of the white area (EDAX). The Fe and Si are the main

impurities as a residual ferrosilicon.

A B C

Fig. 6. BSE/SEM image of crown magnesium crystals. A & B) Show the crystalline magnesium covered with the late stage

low crystalline magnesium. C) Late stage low crystalline magnesium deposited on the high crystalline magnesium.

Dow

nloa

ded

from

ijm

se.iu

st.a

c.ir

at 1

1:39

IRS

T o

n W

edne

sday

Feb

ruar

y 19

th 2

020

24

Iranian Journal of Materials Science & Engineering Vol. 8, Number 2, Spring 2011

ACKNOWLEDGMENT

Authors would like to thank the Iranian

Ministry of Industries and Mines for financial

support.

REFERENCES

1. Habashi, F., Handbook of Extractive

Metallurgy, 1997, Vol. 2; Wiley.

2. Cherubini, F., Raugei, M. and Ulgiati, S., “LCA

of magnesium production; technological

overview and worldwide estimation of

environmental burdens”. Resources,

Conservation and Recycling, 2008, 52, 1093-

1100.

3. Ding, W. and Zang, J., “The Pidgeon process in

China”, Proceedings of 3rd Annual Australasian

Magnesium Conference, Sydney, Australia,

2001, 7.

4. Zang, J. and Ding, W., “The Pidgeon process in

China and its future” Magnesium Technology

2001, Proceedings of the Mineral, Metals and

Material Society Annual Meeting, New

Orleans, Louisiana, 2001, 7-10.

5. Mehrabi, B., “The most suitable technology for

magnesium production from Mg-bearing

minerals in Iran”. University of Industries and

Mines, Makta1324-MIM grant, Final Report,

2005 (In Persian).

6. Mehrabi, B. and Masoudi, F., “Magnesium

metal production using Lloyd Pidgeon

process”. University of Industries and Mines,

Makta 17238-MIM grant, Final Report, 2009

(In Persian)

7. Dickson, J. A. D., “A modified technique for

carbonates in thin section”. Nature, 1965, 205,

587.

8. Sibley, D.F., Gregg, J.M., “Classification of

dolomite rock textures”. Journal of Sedimentary

Petrology, 1987, 57, 967–975.

9. Pidgeon, L. M. and Alexander, W. A., “Thermal

production of magnesium-Pilot plant studies on

the retort ferrosilicon process”. Transactions

AIME, 1944, 159, 315-352.

10. Wulandari, W., Rhamdhani, A., Brooks, G. &

Monaghan, B. J., “Distribution of impurities in

magnesium via silicothermic reduction”

Proceedings of European Metallurgical

Conference, Austria, 2009, 1401-1415.

11. Mayer A., “Plant production of magnesium by

the ferrosilicon process”. Proceedings of

AIMM, New York Meeting: reduction and

refining of non-ferrous metals. Trans Am. Inst.

Mining Mater. Eng., 1944; 363–376.

12. Jones, R. T., “Computer Simulation of

Pyrometallurgical Processes”; Proceedings of

the Twentieth International Symposium on the

Application of Mathematics and Computers in

the Minerals Industries, 1987, V2, 265.

13. Abdellatif, M., “Pilot plant demonstration of the

Mintek Thermal Magnesium Process”

Proceedings of the International Symposium on

Magnesium Technology in the Global Age,

Quebec, Canada, 2006, 67-80.

14. Abdellatif, M., “Refining testwork on crude

magnesium produced in the Mintek Thermal

Magnesium Process”, Proceedings of Southern

African Pyrometallurgy International

Conference, Johannesburg, South Africa, 2006,

343-355.

Dow

nloa

ded

from

ijm

se.iu

st.a

c.ir

at 1

1:39

IRS

T o

n W

edne

sday

Feb

ruar

y 19

th 2

020